Change the logic so we do not allocate memory nor map the device
before actually posting to the REG_UMR QP. In addition, unmap and free
the memory after we get completion.
Signed-off-by: NEli Cohen <eli@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

The patch fixes the rollback in case of failure in creating SRQ.
Signed-off-by: NMoshe Lazer <moshel@mellanox.com>
Signed-off-by: NEli Cohen <eli@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Destroying the workqueue without flushing it first can lead to a case
in which the kernel tries to push a delayed work to the workqueue
which does not exist anymore.
Signed-off-by: NMoshe Lazer <moshel@mellanox.com>
Signed-off-by: NEli Cohen <eli@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1. Make sure wqe_cnt does not exceed the limit published by firmware.

2. There is no requirement that the number of outstanding work
   requests will be a power of two. Remove the ilog2 in the
   calculation of sq.max_post to fix that.

3. Add case for IB_QPT_XRC_TGT in sq_overhead and return 0 as XRC
   target QPs do not have a send queue.
Signed-off-by: NEli Cohen <eli@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

After the last architecture switched to generic hard irqs the config
options HAVE_GENERIC_HARDIRQS & GENERIC_HARDIRQS and the related code
for !CONFIG_GENERIC_HARDIRQS can be removed.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix:

   drivers/infiniband/hw/ocrdma/ocrdma_verbs.c: In function 'ocrdma_build_fr':
>> drivers/infiniband/hw/ocrdma/ocrdma_verbs.c:1832:7: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
     mr = (struct ocrdma_mr *)qp->dev->stag_arr[(hdr->lkey >> 8) &
          ^
   drivers/infiniband/hw/ocrdma/ocrdma_verbs.c: In function 'ocrdma_alloc_frmr':
>> drivers/infiniband/hw/ocrdma/ocrdma_verbs.c:2661:64: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
     dev->stag_arr[(mr->hwmr.lkey >> 8) & (OCRDMA_MAX_STAG - 1)] = (u64) mr;
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Commit 36a8f01c ("IB/qib: Add congestion control agent
implementation") caused statements to leak pass the header guard.
Fix this.
Reviewed-by: NMike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: NIra Weiny <ira.weiny@intel.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Fix passing wrong opcode to ocrdma_modify_srq and query SRQ.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

In UMC case, driver needs to fill PVID in the address vector
template for UD traffic.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Add ABI versioning support between driver and userspace library.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

While posting inline DPP data, we are not considering multiple sges.
Fix this.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1) Increase STAG Array size.
2) Max inline data size should be set to the same value
   used during QP creation
3) Set max_sge_rd to zero since we dont support RD transport in our adapters.
4) Max cqes reported in ibv_devinfo should be from QUERY_CONFIG.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Create_CQ verb doesn't provide a PD pointer.  So, until now we are
creating all (both userspace and kernel) CQ DB regions from PD0.  This
will result in mmapping PD0 to applications.  A rogue userspace
application can mess things up.

Also more serious issues is even the be2net NIC uses PD0.

This patch addresses this problem by:

1) Create a PD page for every userspace application when the
   alloc_ucontext is called. This will be destroyed in
   dealloc_ucontext.
2) All CQs for that context will use the PD allocated in ucontext.
3) The first create_PD call from application will result in returning
   the PD address from its ucontext (no new PD will be created).
4) For subsecquent create_pd calls from application, we create new PDs for
   the application.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1) Fixed setting FR_MR bit for FRWR stag allocation
2) Access rights are passsed during FRWR stage and not during STAT allocation stage
3) FRWR WQE structure cleanup
4) Add QP level signaled bit.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1) All RQ doorbells are handled by ERX2 and doorbell->num_posted
   offset is constant to bit offset 24 for ERX2 irrspective of Q id.

2) Fixed RESET to INIT state change (from ERR->RST->INIT->RTR case).
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

There are cases like SRIOV where can get only one MSI-X vector
allocated for RoCE.  In that case we need to use the vector for both
data plane and control plane.  We need to use EQ create version V2.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Also increase MAX AH to 512.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Also get the max_srq value from query_config mailbox response.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1)	Fix ocrdma_get_num_posted_shift for upto 128 QPs.
2)	Create for min of dev->max_wqe and requested wqe in create_qp.
3)	As part of creating ird queue, populate with basic header templates.
4)	Make sure all the DB memory allocated to userspace are page aligned.
5)	Fix issue in checking the mmap local cache.
6)	Some code cleanup.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Implement ib_create_flow() and ib_destroy_flow().

Translate the verbs structures provided by the user to HW structures
and call the MLX4_QP_FLOW_STEERING_ATTACH/DETACH firmware commands.

On the ATTACH command completion, the firmware provides a 64-bit
registration ID, which is placed into struct mlx4_ib_flow that wraps
the instance of struct ib_flow which is retuned to caller.  Later,
this reg ID is used for detaching that flow from the firmware.
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Don't emit OOM warnings when k.alloc calls fail when
there there is a v.alloc immediately afterwards.

Converted a kmalloc/vmalloc with memset to kzalloc/vzalloc.
Signed-off-by: NJoe Perches <joe@perches.com>
Acked-by: N"Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Lustre uses a advertised max MR size of ~0ULL to indicate it should
use a dma_mr.  Hence advertise max MR size as ~0ULL.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

When polling, we do a GTS update if the accumulated cidx_inc == the CQ
depth / 16.  However, if the CQ is large enough, Cq depth / 16 exceeds
the size of the field in the GTS word.  So we also need to update if
cidx_inc hits CIDXINC_MASK to avoid overflowing the field.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

accept_cr() failed to set the arp error handler on a reused skb.  This
results in a kernel crash if the arp does indeed time out.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

When determining how many WRs are completed with a signaled CQE,
correctly deal with queue wraps.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This patch makes following fixes in QP flush logic:

- correctly flushes unsignaled WRs followed by a signaled WR
- supports for flushing a CQ bound to multiple QPs
- resets cidx_flush if a active queue starts getting HW CQEs again
- marks WQ in error when we leave RTS. This was only being done for
  user queues, but we need it for kernel queues too so that
  post_send/post_recv will start returning the appropriate error
  synchronously
- eats unsignaled read resp CQEs. HW always inserts CQEs so we must
  silently discard them if the read work request was unsignaled.
- handles QP flushes with pending SW CQEs. The flush and out of order
  completion logic has a bug where if out of order completions are
  flushed but not yet polled by the consumer and the qp is then
  flushed then we end up inserting duplicate completions.
- c4iw_flush_sq() should only flush wrs that have not already been
  flushed.  Since we already track where in the SQ we've flushed via
  sq.cidx_flush, just start at that point and flush any remaining.
  This bug only caused a problem in the presence of unsignaled work
  requests.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>

[ Fixed sparse warning due to htonl/ntohl confusion.  - Roland ]
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Move QP to TERMINATE instead to allow the peer to get the TERM
message. This bug wasn't detectable until newer FW that moves
connections out of RDMA mode as soon as an error is detected.

QP can exit RTS before the last AE arrives.  This was introduced by
changes in the FW to kick connections out of RDMA mode as soon as an
error is detected.  A side effect of this is that the driver can move
the QP out of RTS before the AE causing the connection to get kicked
out of RDMA mode is processed.  Fix for this is to always post async
errors even if the QP is out of RTS.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NVipul Pandya <vipul@chelsio.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Add new cpl messages, cpl_act_open_req6 and cpl_t5_act_open_req6, for
initiating active open connections.

Use LLD api cxgb4_create_server and cxgb4_create_server6 for
initiating passive open connections. Similarly use cxgb4_remove_server
to remove the passive open connections in place of listen_stop.

Add support for iWARP over VLAN device and enable IPv6 support on VLAN device.

Make use of import_ep in c4iw_reconnect.
Signed-off-by: NVipul Pandya <vipul@chelsio.com>

[ Fix build when IPv6 is disabled and make sure iw_cxgb4 is not built-in
  when ipv6 is a module.  - Roland ]
Signed-off-by: NRoland Dreier <roland@purestorage.com>

PCI core will initialize device MSI/MSI-X capability in
pci_msi_init_pci_dev().  So device drivers should use
pci_dev->msi_cap/msix_cap to determine whether a device supports
MSI/MSI-X instead of using pci_find_capability(pci_dev,
PCI_CAP_ID_MSI/MSIX).  Access to PCIe device config space again will
consume more time.
Signed-off-by: NYijing Wang <wangyijing@huawei.com>
Acked-by: NMike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

struct pci_driver qib_driver is only used in qib_init.c.  Remove it
from qib.h and make it static in qib_init.c.
Signed-off-by: NPaul Bolle <pebolle@tiscali.nl>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1. The code accepts chunks of messages, and splits the chunk into
   packets when converting packets into sdma queue entries.  Adjacent
   packets will use user buffer pages smartly to avoid pinning the
   same page multiple times.

2. Instead of discarding all the work when SDMA queue is full, the
   work is saved in a pending queue.  Whenever there are enough SDMA
   queue free entries, pending queue is directly put onto SDMA queue.

3. An interrupt handler is used to progress this pending queue.
Reviewed-by: NMike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: NCQ Tang <cq.tang@intel.com>
Signed-off-by: NMike Marciniszyn <mike.marciniszyn@intel.com>

[ Fixed up sparse warnings.  - Roland ]
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Modify the type of local_addr and remote_addr fields in struct
iw_cm_id from struct sockaddr_in to struct sockaddr_storage to hold
IPv6 and IPv4 addresses uniformly.

Change the references of local_addr and remote_addr in cxgb4, cxgb3,
nes and amso drivers to match this.  However to be able to actully run
traffic over IPv6, low-level drivers have to add code to support this.
Signed-off-by: NSteve Wise <swise@opengridcomputing.com>
Reviewed-by: NSean Hefty <sean.hefty@intel.com>

[ Fix unused variable warnings when INFINIBAND_NES_DEBUG not set.
  - Roland ]
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1) In post recv, don't ring the DB doorbell if the QP is in RTR state.
   Cache the DB calls, until the QP is moved to RTS state.
2) Add max_rd_sge support to dev->attr.
3) Code cleanup in alloc_pd path.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

1) Remove __packed for structures.
2) Align and pad all ABI structure to 64 bit boundaries
   instead of using __packed.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Remove QP state machine in ocrdma low-level driver and use on the core
IB stack's instead.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Remove redundant dev reference from structures:

1) ocrdma_cq.
2) ocrdma_ah.
3) ocrdma_hw_mr.
4) ocrdma_mw.
5) ocrdma_srq.
Signed-off-by: NNaresh Gottumukkala <bgottumukkala@emulex.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>